Intelligent learning behavior analysis and knowledge mastery evaluation method

Abstract

The present invention relates to the field of intelligent education technology, and specifically to an intelligent learning behavior analysis and knowledge mastery evaluation method, comprising the following steps: S1: collecting learners' behavior data during the learning process; S2: performing data cleaning and data normalization processing on the behavior data collected in S1 to form a behavior data set; S3: extracting the behavior characteristics of learners at each learning stage from the behavior data set; S4: using a cluster analysis algorithm to build a learner's behavior pattern model to identify the behavior patterns of different types of learners; S5: evaluating the learner's knowledge mastery at each knowledge point; S6: generating personalized learning suggestions based on the evaluation results of the knowledge mastery in S5. The present invention, through in-depth analysis of learners' behavior data, realizes accurate knowledge mastery evaluation and personalized learning suggestion generation, can optimize learning effects, and improve learners' learning efficiency and knowledge mastery level.

Description

Technical Field

[0001] The present invention relates to the field of intelligent education technology, and in particular to an intelligent learning behavior analysis and knowledge mastery evaluation method. Background Art

[0002] With the development of information technology, intelligent learning systems have become an important tool in the field of education; traditional learning assessment methods mainly rely on final exams or regular assessments, which usually cannot timely and accurately reflect the learners' true status and progress in the learning process; with the diversification of learning content, a single assessment method cannot fully understand students' mastery of various knowledge points; especially in the context of the growing demand for personalized learning and adaptive learning, how to track learners' learning progress in real time, accurately assess their knowledge mastery, and provide personalized learning suggestions based on this has become an important issue facing current educational technology.

[0003] However, the intelligent learning systems in the existing technology still have many limitations in learning behavior analysis and knowledge mastery assessment. Most of the existing systems rely on traditional behavior monitoring methods, such as time, answer accuracy, etc., and lack in-depth analysis of learners' behavior characteristics, especially in how to comprehensively consider learners' learning progress, error types, and learning habits. At the same time, the existing technology relies on relatively simple rules or static data analysis to generate personalized learning suggestions, and cannot be effectively adjusted according to learners' dynamic learning behaviors and personalized needs. Therefore, there is an urgent need for an intelligent learning behavior analysis and knowledge mastery assessment method to solve the above problems. Summary of the invention

[0004] Based on the above objectives, the present invention provides a method for intelligent learning behavior analysis and knowledge mastery assessment.

[0005] The intelligent learning behavior analysis and knowledge mastery evaluation method includes the following steps:

[0006] S1: Collect learners’ behavioral data during the learning process, including learning time, learning frequency, learning progress, answering speed, error rate, and click behavior;

[0007] S2: Perform data cleaning and data normalization on the behavioral data collected in S1 to form a behavioral data set;

[0008] S3: Extract the behavioral characteristics of learners at each learning stage from the behavioral dataset, including learners’ activity, learning time distribution, and answer accuracy;

[0009] S4: Based on the behavioral characteristics of the learners, a cluster analysis algorithm is used to construct a behavior pattern model of the learners, so as to identify the behavior patterns of different types of learners;

[0010] S5: Based on the identified behavior patterns and the learners’ learning performance on each knowledge point, the learners’ knowledge mastery on each knowledge point is evaluated. The specific evaluation includes the progress of knowledge point mastery, error type analysis and review frequency.

[0011] S6: Generate personalized learning suggestions based on the evaluation results of knowledge mastery in S5, including learning progress adjustment, wrong question review and review content recommendation.

[0012] Optionally, the S1 specifically includes:

[0013] S11: collecting the learner's learning behavior data in real time through a sensor module arranged on a learning device or a learning platform, wherein the sensor module includes a user input device, a display device and an interaction submodule, and is used to record the learner's interaction behavior during the learning process;

[0014] S12: Based on the learners' interactive behavior data on the learning platform, record the learners' learning time and learning frequency. The learning time is calculated by recording the time difference between each learner's login and exit from the learning platform, and the learning frequency is calculated by recording the time interval between each learner's learning.

[0015] S13: by analyzing the learning tasks selected by the learners in the learning platform, recording the learning progress, wherein the learning progress is obtained by calculating the time taken by the learners to complete the tasks at each stage and the task completion status;

[0016] S14: by monitoring the learner's input response during the process of answering questions, the learner's answering speed is recorded, and the answering speed is counted by calculating the time required for the learner to answer each question;

[0017] S15: recording the learner's error rate according to the errors made during the learner's answering process, wherein the error rate is calculated by calculating the ratio of the number of wrong answers to the total number of answers;

[0018] S16: By analyzing the learner's click behavior, the learner's click behavior data on the platform is recorded, and the click behavior data includes the page clicked by the learner, the selected learning content and the browsing time.

[0019] Optionally, the S2 specifically includes:

[0020] S21: Data cleaning is performed on the behavior data collected in S1, wherein the data cleaning includes removing missing values, removing outliers, and removing duplicate data; wherein, the processing of missing values ​​adopts the mean filling method; outliers are identified by Z-Score detection based on statistical methods, and if the Z-Score value of a data point exceeds a preset threshold, it is regarded as an outlier and is removed; duplicate data is screened by comparing timestamps and duplicates are removed;

[0021] S22: Perform data normalization on the cleaned behavioral data to ensure that each data has the same dimension;

[0022] S23: Classify the cleaned and normalized behavior data according to the learning stage, learning task and task completion status, and store them in a data set library to form a behavior data set.

[0023] Optionally, the S3 specifically includes:

[0024] S31: According to the behavioral data set processed in S2, the learner's learning activities are processed in stages, and the learning process is divided into several learning stages, where the learning stages are defined according to different progress of the learning task or learning time interval;

[0025] S32: Based on the behavior data of each divided learning stage, the activity of the learner in the stage is calculated, and the activity is obtained by counting the ratio of the learner's effective learning time to the total learning time in the stage;

[0026] S33: Calculate the distribution of learners’ learning time in each learning stage. The distribution of learning time is measured by calculating the proportion of learning time in each learning stage to the total learning time.

[0027] S34: extracting the correctness of the learner's answers in each learning stage, wherein the correctness of the answers is obtained by calculating the ratio of the number of answers completed by the learner in that stage to the number of correct answers.

[0028] Optionally, the S31 specifically includes:

[0029] S311: According to the progress of the learning task, the learning process is divided into several stages, each stage corresponds to a different progress point of the learning task, and the progress point is defined according to the proportion of the learner completing the task. The specific calculation formula is: Among them, P task is the completion progress of the learning task, T completed is the amount of tasks completed by the learner, T total is the total workload of the task; if P task If it is greater than the preset threshold, it is divided into a new learning stage;

[0030] S312: Divide the learning process into time periods according to the learning time intervals of the learners in the learning process, and each time period is represented as a learning stage; the learning time intervals are divided by setting a fixed time window, and the time window size is the average learning time of the learners. The specific calculation formula is: Among them, T window is the time window size for each learning stage, T total is the total learning time of the learner, N windows The number of learning stages to be set;

[0031] S313: According to the progress of the above learning task and the result of the learning time interval division, determine the start time and end time of each learning stage. The start time and end time are determined by the intersection of the corresponding task progress and the learning time window. The specific calculation formula is: start =T previous +T window and T end =T start +T window , where T start and T end are the start time and end time of the current learning phase, T previous is the end time of the previous learning phase, T window is the time window for each learning phase.

[0032] Optionally, the S4 specifically includes:

[0033] S41: Select the K-means clustering algorithm to perform cluster analysis on the learner's behavior characteristics. By setting the number of categories K and initializing K center points, the distance between each learner and the center point is calculated, and the learner is assigned to the cluster with the closest distance. This process is repeated until the clustering results converge.

[0034] S42: Standardize the learners’ behavioral characteristics by using the mean-variance standardization method, adjusting the mean of each characteristic to 0 and the standard deviation to 1, to ensure that each characteristic has equal weight in the cluster analysis;

[0035] S43: Through the iterative process of the K-means clustering algorithm, the position of the cluster center point is gradually optimized until the clustering results converge, and each cluster represents the behavior pattern of a learner;

[0036] S44: Build a learner behavior pattern model based on the clustering results. The center point of each cluster represents a behavior pattern, which is used to describe the typical behavior characteristics of learners in the cluster, including learning activity, learning time distribution or answer accuracy rate;

[0037] S45: Based on the learner behavior pattern model, different types of learners are identified by analyzing the characteristic values ​​of the clusters.

[0038] Optionally, the S41 specifically includes:

[0039] S411: first set the number of categories K, and randomly select K cluster centers, wherein the number of categories K is preset according to the characteristics of the learning task and the distribution of the data;

[0040] S412: Calculate the distance between each learner and the K cluster centers;

[0041] S413: assigning each learner to the cluster closest to it according to the calculated distance, assigning the learner to the cluster with the smallest distance to it, and updating the cluster affiliation;

[0042] S414: Update the center point position of the cluster by calculating the mean of the behavioral characteristics of all learners in each cluster to update the center point of the cluster;

[0043] S415: Repeat S412 and S413 until the clustering result converges, that is, the cluster center point no longer changes significantly, or the preset maximum number of iterations is reached;

[0044] S416: Through the above clustering process, learners are divided into several clusters according to their behavioral characteristics, and each cluster represents a behavioral pattern of a learner.

[0045] Optionally, the S45 specifically includes:

[0046] S451: According to the learner behavior pattern model constructed in S44, the characteristic value of each cluster is obtained, including learning activity, learning time distribution and answer correctness; wherein the learning activity is calculated by the following formula: Among them, A k is the learning activity of the kth cluster, T k is the total learning time of learners in the cluster, T total is the total learning time of all learners;

[0047] S452: Classify learners into different types according to the learning activity values ​​of the clusters; if the activity of learners in a cluster is higher than a predetermined threshold, the cluster represents high-activity learners; if it is lower than the predetermined threshold, the cluster represents low-activity learners;

[0048] S453: Analyze the learning progress of the learners according to the learning time distribution of the cluster, and calculate the learning progress of the learners in the cluster by the following formula: Among them, P k is the learning progress of the kth cluster, T current is the learning time of the current stage, T totalis the total learning time of the learner from the beginning to the present; if P k If it is greater than a predetermined threshold, the corresponding cluster is identified as a fast progress learner;

[0049] S454: Based on the learner's performance in answering questions correctly, the learner type is refined; the answering correctness feature is calculated by the following formula: Among them, P correct is the correctness of the answer of the kth cluster, C k is the number of correct answers given by learners in the cluster, Q k is the total number of answers of learners in the cluster; if P correct If it is higher than the set threshold, the corresponding family represents high-accuracy learners, and if it is lower than the set threshold, it represents low-accuracy learners;

[0050] S455: Through steps S451-S454, the activity, learning progress and answer accuracy of the learners in the cluster are comprehensively analyzed, and the learner types are divided into high-activity learners, low-activity learners, fast-progress learners, slow-progress learners, high-accuracy learners or low-accuracy learners.

[0051] Optionally, the S5 specifically includes:

[0052] S51: According to the learner behavior pattern identified in S45, the learning performance data of each learner on each knowledge point is obtained, and the learning performance is defined as the correct answer rate of each knowledge point;

[0053] S52: Evaluate the learner's progress in mastering each knowledge point based on the correct rate of the learner's answers to each knowledge point; calculate using the following formula: Among them, P i,j is the progress of the i-th learner in mastering the j-th knowledge point, C i,j is the correct answer rate of the i-th learner on the j-th knowledge point, T j is the maximum score of this knowledge point; if P i,j If the value is greater than the predetermined threshold, it means that the learner has mastered the knowledge point. If it is lower than the predetermined threshold, it means that the learner has not yet mastered the knowledge point.

[0054] S53: According to the number of wrong answers of each learner on each knowledge point, error type analysis is conducted. Error types include selection errors, calculation errors or concept comprehension errors. Specifically, error type analysis is determined by calculating the proportion of each type of error. If the proportion of selection errors is greater than 30%, it is considered that the learner of this knowledge point has difficulty with this type of error and needs to focus on reviewing.

[0055] S54: Calculate the learner's review frequency based on the learner's mastery progress on each knowledge point and the error type analysis results; the review frequency is determined by the learner's mastery of a certain knowledge point and the error type analysis results; specifically, if the learner's mastery progress on a certain knowledge point is lower than a preset value, and the error type analysis results show that the error ratio is greater than 30%, then increase the review frequency of the corresponding knowledge point.

[0056] Optionally, the S6 specifically includes:

[0057] S61: Adjust the learner's learning progress according to the analysis results of each learner's mastery progress and error types on each knowledge point; specifically, if the learner's mastery progress on a certain knowledge point is lower than a predetermined threshold, it will be recommended to increase the learning time of the corresponding knowledge point or arrange additional learning tasks; the learning progress adjustment formula is:

[0058] S62: Generate wrong question review suggestions based on the error type analysis results in S53; the wrong question review suggestions are generated by the following steps:

[0059] Step 1: Sort by error type, reviewing high-frequency error types first;

[0060] Step 2: Recommend corresponding exercises or analysis videos for the most common error types;

[0061] S63: Based on the learner's progress in mastering each knowledge point and the analysis of the error types, generate review content recommendations, including review questions, review chapters or related explanation videos.

[0062] Beneficial effects of the present invention:

[0063] The present invention, by comprehensively analyzing the behavioral data of learners in the learning process, accurately evaluates their mastery of various knowledge points, and can provide learners with accurate learning progress adjustment, wrong question review and review content recommendation; this evaluation method based on learners' dynamic behavior solves the problem that traditional evaluation methods cannot track learners' learning status in real time, can timely discover learners' weak links in the learning process, and give effective feedback and guidance, greatly improving learning efficiency.

[0064] The present invention, by adopting a cluster analysis algorithm to construct a learner behavior pattern model, can identify different types of learners and provide customized learning suggestions based on their personalized learning characteristics; this personalized learning path and review strategy can adjust the learning plan in a targeted manner, optimize the learning content recommendation, and thus help learners master knowledge points more efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0066] Figure 1 A schematic diagram of a method for evaluating knowledge mastery according to an embodiment of the present invention;

[0067] Figure 2 Schematic diagram of a method for constructing a learner's behavior pattern model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0068] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0069] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0070] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0071] like Figure 1-Figure 2 As shown, the intelligent learning behavior analysis and knowledge mastery evaluation method includes the following steps:

[0072] S1: Collect learners’ behavioral data during the learning process, including learning time, learning frequency, learning progress, answering speed, error rate, and click behavior;

[0073] S2: Perform data cleaning and data normalization on the behavioral data collected in S1 to form a behavioral data set;

[0074] S3: Extract the behavioral characteristics of learners at each learning stage from the behavioral dataset, including learners’ activity, learning time distribution, and answer accuracy;

[0075] S4: Based on the behavioral characteristics of learners, a cluster analysis algorithm is used to build a learner behavior pattern model to identify the behavior patterns of different types of learners;

[0076] S5: Based on the identified behavior patterns and the learners’ learning performance on each knowledge point, the learners’ knowledge mastery on each knowledge point is evaluated. The specific evaluation includes the progress of knowledge point mastery, error type analysis and review frequency.

[0077] S6: Generate personalized learning suggestions based on the evaluation results of knowledge mastery in S5, including learning progress adjustment, wrong question review and review content recommendation.

[0078] S1 specifically includes:

[0079] S11: collecting the learner's learning behavior data in real time through a sensor module set on the learning device or learning platform, where the sensor module includes a user input device, a display device and an interaction submodule, and is used to record the learner's interaction behavior during the learning process;

[0080] S12: Based on the learners' interactive behavior data on the learning platform, record the learners' learning time and learning frequency. The learning time is calculated by recording the time difference between each learner's login and exit from the learning platform, and the learning frequency is calculated by recording the time interval between each learner's learning.

[0081] S13: By analyzing the learning tasks selected by the learners in the learning platform, the learning progress is recorded. The learning progress is obtained by calculating the time taken by the learners to complete the tasks at each stage and the completion status of the tasks. Specifically, the calculation formula of the learning progress is: Among them, P is the learning progress, T completed is the time the learner has completed the task, T total The total task time allocated to the learner; if the task is staged, the progress of each stage is accumulated according to the completion of each stage to get the total learning progress;

[0082] S14: by monitoring the learner's input response during the process of answering questions, the learner's answering speed is recorded, and the answering speed is counted by calculating the time required for the learner to answer each question;

[0083] S15: according to the errors made by the learners in answering questions, the learners' error rate is recorded, and the error rate is calculated by calculating the ratio of the number of wrong answers to the total number of answers;

[0084] S16: By analyzing the learners' click behavior, the learners' click behavior data on the platform are recorded. The click behavior data includes the pages clicked by the learners, the learning content selected and the browsing time. The click data in the learning process is recorded and analyzed in real time using click stream technology. Through the above steps, various types of learners' behavior data can be collected comprehensively and accurately, and specific technical means such as sensor modules, click stream technology, etc. can be used to ensure the accuracy and real-time nature of the data, so as to provide high-quality data support for the subsequent extraction of behavioral features and the construction of learners' behavior patterns, and then provide learners with personalized learning suggestions and improve learning effects.

[0085] S2 specifically includes:

[0086] S21: Data cleaning is performed on the behavioral data collected in S1. Data cleaning includes removing missing values, removing outliers, and removing duplicate data. Missing values ​​are processed by the mean filling method. Outliers are identified by Z-Score detection based on statistical methods. If the Z-Score value of a data point exceeds a preset threshold, it is considered an outlier and is removed. Duplicate data is screened by comparing timestamps and duplicates are removed.

[0087] S22: Perform data normalization on the cleaned behavioral data to ensure that each data has the same dimension; data normalization uses the minimum-maximum standardization method, and the formula is as follows: Among them, X′ is the normalized data, X is the original data, and X min and X max are the minimum and maximum values ​​in the original data respectively; the normalized data range is compressed to the [0, 1] interval for subsequent processing;

[0088] S23: The cleaned and normalized behavioral data are classified according to the learning stage, learning tasks and task completion status, and stored in the data set library to form a behavioral data set; the behavioral data set is indexed according to the learner's learning path and behavioral characteristics to facilitate subsequent rapid retrieval and analysis; through the above steps, it can be ensured that the collected behavioral data is effectively cleaned and standardized in the preprocessing stage, noise is removed and the dimensions are unified to ensure the accuracy and consistency of the data; these technical means help to provide high-quality data support for subsequent learner behavior pattern recognition and knowledge mastery assessment, and thus provide a reliable basis for the generation of personalized learning recommendations.

[0089] S3 specifically includes:

[0090] S31: Based on the behavioral data set processed in S2, the learner's learning activities are processed in stages, and the learning process is divided into several learning stages. The learning stages are defined according to different progress of the learning task or learning time intervals, such as the course chapters or the learner's task completion progress.

[0091] S32: Based on the behavioral data of each divided learning stage, the activity of the learner in that stage is calculated. The activity is calculated by counting the ratio of the learner's effective learning time to the total learning time in that stage. The specific calculation formula is: Among them, A is the activity, T active is the effective learning time of learners in the corresponding stage, T total is the total learning time of the corresponding stage. If the learner's effective learning time in this stage exceeds 50% of the total time, the stage is considered active;

[0092] S33: Calculate the distribution of the learner's learning time in each learning stage. The distribution of learning time is measured by calculating the proportion of the learning time in each learning stage to the total learning time. The specific calculation formula is: Where D is the learning time distribution, T stage is the learning time of a certain learning stage, T total is the total learning time of the learners;

[0093] S34: extract the correctness of the learner's answers in each learning stage. The correctness of the answers is calculated by calculating the ratio of the number of answers completed by the learner in this stage to the number of correct answers. The specific calculation formula is: Among them, C is the correctness of the answer, N correct is the number of questions answered correctly by the learner in this stage, N total is the total number of questions completed by the learners in this stage; through the above steps, the behavioral characteristics of the learners in each learning stage can be accurately extracted from the behavioral data set, including activity, learning time distribution and answer accuracy. By calculating the activity and learning time distribution, the learners' participation and learning efficiency can be comprehensively measured; through the correctness of the answers, the learners' knowledge mastery at each stage can be quantified. These behavioral characteristics provide detailed data support for the subsequent learners' behavior pattern recognition and knowledge mastery evaluation.

[0094] S31 specifically includes:

[0095] S311: According to the progress of the learning task, the learning process is divided into several stages. Each stage corresponds to a different progress point of the learning task. The progress point is defined according to the proportion of the learner completing the task. The specific calculation formula is: Among them, P task is the completion progress of the learning task, T completedis the amount of tasks completed by the learner, T total is the total workload of the task; if P task If it is greater than the preset threshold, it is divided into a new learning stage;

[0096] S312: Divide the learning process into time periods according to the learning time intervals of the learners in the learning process, and each time period is represented as a learning stage; the learning time intervals are divided by setting a fixed time window, and the time window size is the average learning time of the learners. The specific calculation formula is: Among them, T window is the time window size for each learning stage, T total is the total learning time of the learner, N windows The number of learning stages to be set;

[0097] S313: According to the progress of the above learning task and the result of the learning time interval division, determine the start time and end time of each learning stage. The start time and end time are determined by the intersection of the corresponding task progress and the learning time window. The specific calculation formula is: T start =T previous +T window and T end =T start +T window , where T start and T end are the start time and end time of the current learning phase, T previous is the end time of the previous learning phase, T window is the time window for each learning stage; through the above steps, the various learning stages in the learning process can be accurately divided according to the progress of the learning task and the learner's learning time interval; this division method can not only dynamically adjust the length of the learning stage based on the learner's actual progress, but also ensure that the data of each learning stage is more in line with the learner's actual learning situation, providing more accurate and appropriate data support for subsequent behavioral feature extraction, knowledge mastery assessment and personalized learning suggestion generation.

[0098] S4 specifically includes:

[0099] S41: Select K-means clustering algorithm to perform cluster analysis on learners' behavioral characteristics. By setting the number of categories K and initializing K center points, the distance between each learner and the center point is calculated, and the learners are assigned to the cluster with the closest distance. This process is repeated until the clustering results converge, thereby dividing the learners into several behavioral pattern clusters.

[0100] S42: Standardize the learners’ behavioral characteristics by using the mean-variance standardization method, adjusting the mean of each characteristic to 0 and the standard deviation to 1, to ensure that each characteristic has equal weight in the cluster analysis;

[0101] S43: Through the iterative process of the K-means clustering algorithm, the position of the cluster center point is gradually optimized until the clustering results converge, and each cluster represents the behavior pattern of a learner;

[0102] S44: Build a learner behavior pattern model based on the clustering results. The center point of each cluster represents a behavior pattern, which is used to describe the typical behavior characteristics of learners in the cluster, including learning activity, learning time distribution or answer accuracy rate;

[0103] S45: Based on the learner behavior pattern model, different types of learners can be identified by analyzing the characteristic values ​​of the clusters, providing a basis for subsequent learning progress adjustments and knowledge mastery assessment.

[0104] S41 specifically includes:

[0105] S411: Select the K-means clustering algorithm to perform cluster analysis on the learner's behavior characteristics. First, set the number of categories K and randomly select K cluster centers. The number of categories K is preset according to the characteristics of the learning task and the distribution of the data.

[0106] S412: Calculate the distance between each learner and the K cluster centers. The distance is calculated using the Euclidean distance formula. The Euclidean distance formula is as follows: Where D is the distance between the learner and the cluster center, x i is the behavioral characteristic value of the learner, c i is the eigenvalue of the cluster center, and n is the dimension of the behavior feature;

[0107] S413: assigning each learner to the cluster closest to it according to the calculated distance, assigning the learner to the cluster with the smallest distance to it, and updating the cluster affiliation;

[0108] S414: Update the center point of the cluster by calculating the mean of the behavioral characteristics of all learners in each cluster. The new cluster center point is updated by the following formula: Among them, C k is the center point of the kth cluster, N k is the number of learners in cluster k, x i is the behavioral characteristics of the i-th learner in the cluster;

[0109] S415: Repeat S412 and S413 until the clustering result converges, that is, the cluster center point no longer changes significantly, or the preset maximum number of iterations is reached;

[0110] S416: Through the above clustering process, learners are divided into several clusters according to their behavioral characteristics, and each cluster represents a behavioral pattern of a learner; learners in each cluster have similar behavioral characteristics, and the behaviors between clusters are quite different, which provides data support for subsequent learning progress adjustment and knowledge mastery evaluation; through the above steps, learners' behavioral patterns can be effectively clustered and analyzed, and learners can be accurately divided into different behavioral pattern clusters; this process ensures the accuracy of the behavioral pattern through Euclidean distance calculation and mean update mechanism, which provides strong data support for personalized learning path recommendation and subsequent knowledge mastery evaluation.

[0111] S45 identifies different types of learners, including:

[0112] S451: According to the learner behavior pattern model constructed in S44, the characteristic value of each cluster is obtained, including learning activity, learning time distribution and answer correctness; wherein the learning activity is calculated by the following formula: Among them, A k is the learning activity of the kth cluster, T k is the total learning time of learners in the cluster, T total is the total learning time of all learners;

[0113] S452: Divide learners into different types according to the learning activity value of the cluster; if the activity of learners in the cluster is higher than a predetermined threshold, the cluster represents high-activity learners; if it is lower than the predetermined threshold, it represents low-activity learners; the activity threshold is set according to historical learning data and can be adjusted dynamically; for example, if A k >0. is considered a highly active learner. k <0.3 is considered as a low-activity learner;

[0114] S453: Analyze the learning progress of the learners according to the learning time distribution of the cluster, and calculate the learning progress of the learners in the cluster by the following formula: Among them, P k is the learning progress of the kth cluster, T current is the learning time of the current stage, T total is the total learning time of the learner from the beginning to the present; if P k If it is greater than a predetermined threshold, the corresponding cluster is identified as a fast progress learner;

[0115] S454: Based on the learner's performance in answering questions correctly, the learner type is refined; the answering correctness feature is calculated by the following formula: Among them, P correct is the correctness of the answer of the kth cluster, C kis the number of correct answers given by learners in the cluster, Q k is the total number of answers of learners in the cluster; if P correct If it is higher than the set threshold, the corresponding family represents high-accuracy learners, and if it is lower than the set threshold, it represents low-accuracy learners;

[0116] S455: Through steps S451-S454, the activity, learning progress and answer accuracy of learners in the cluster are comprehensively analyzed, and the learner types are divided into high-activity learners, low-activity learners, fast-progress learners, slow-progress learners, high-accuracy learners or low-accuracy learners; each type of learner has similar behavioral characteristics, and personalized learning suggestions can be provided based on their behavioral characteristics; through the above steps, different types of learners can be accurately analyzed and identified based on the learners' behavioral pattern models, and through multi-dimensional feature analysis such as activity, learning progress and answer accuracy, not only can the learner types be effectively distinguished, but also a basis can be provided for personalized learning suggestions, learning progress adjustments and knowledge mastery evaluation.

[0117] S5 specifically includes:

[0118] S51: According to the learner behavior pattern identified in S45, the learning performance data of each learner on each knowledge point is obtained. The learning performance is defined as the correct answer rate of each knowledge point. The correct answer rate is calculated by the following formula: Among them, C i,j represents the correct answer rate of the i-th learner on the j-th knowledge point, R i,j is the number of correct answers given by the i-th learner at the j-th knowledge point, Q i,j is the total number of answers given by the j-th learner at the j-th knowledge point;

[0119] S52: Evaluate the learner's progress in mastering each knowledge point based on the learner's correct answer rate on each knowledge point; specifically, the learner's progress in mastering a certain knowledge point is calculated using the following formula: Among them, P i,j is the progress of the i-th learner in mastering the j-th knowledge point, C i,j is the correct answer rate of the i-th learner on the j-th knowledge point, T j is the maximum score of this knowledge point; if P i,j If the value is greater than the predetermined threshold, it means that the learner has mastered the knowledge point. If it is lower than the predetermined threshold, it means that the learner has not yet mastered the knowledge point.

[0120] S53: According to the number of wrong answers of each learner on each knowledge point, error type analysis is performed. Error types include selection errors, calculation errors or concept understanding errors. Specifically, error type analysis is determined by calculating the proportion of each type of error. If the proportion of selection errors is greater than 30%, it is considered that the learner of this knowledge point has difficulty with this type of error and needs to focus on reviewing. The error type analysis formula is: Among them, E i,j,k is the proportion of the kth type of error made by the i-th learner on the j-th knowledge point, O i,j,k is the number of type k errors made by the i-th learner on the j-th knowledge point, Q i,j is the total number of questions answered by the learner on the jth knowledge point;

[0121] S54: Calculate the learner's review frequency based on the learner's mastery of each knowledge point and the error type analysis results; the review frequency is determined by the learner's mastery of a certain knowledge point and the error type analysis results; specifically, if the learner's mastery of a certain knowledge point is lower than the preset value, and the error type analysis results show that the error ratio is greater than 30%, then increase the review frequency of the corresponding knowledge point; the formula for the review frequency is: F i,j =R×(1-P i,j )×(1+E i,j,k ), where F i,j is the review frequency of the i-th learner on the j-th knowledge point, R is the constant coefficient, P i,j is the mastering progress of the i-th learner on the j-th knowledge point, E i,j,k is the proportion of error types of the i-th learner on the j-th knowledge point. Through the above steps, the learner's knowledge mastery of each knowledge point can be comprehensively evaluated.

[0122] S6 specifically includes:

[0123] S61: Adjust the learner's learning progress according to the analysis results of each learner's mastery progress and error types on each knowledge point; specifically, if the learner's mastery progress on a certain knowledge point is lower than a predetermined threshold, it will be recommended to increase the learning time of the corresponding knowledge point or arrange additional learning tasks; the learning progress adjustment formula is:

[0124] S62: Based on the error type analysis results in S53, generate wrong question review suggestions; if the learner has many types of errors in certain knowledge points, especially wrong choices or wrong understandings, the system will suggest that they review the wrong questions and conduct targeted review; the strategy for wrong question review includes providing the learner with analysis of all wrong answers to help them master the correct solution methods; wrong question review suggestions are generated through the following steps:

[0125] Step 1: Sort by error type, reviewing high-frequency error types first;

[0126] Step 2: Recommend corresponding exercises or analysis videos for the most common error types;

[0127] S63: Generate review content recommendations based on the learner's progress in mastering each knowledge point and the analysis of the types of errors; the basis for review content recommendations is that the learner has a poor grasp of a certain knowledge point or has a large number of error types. Based on this information, the system provides review materials to the learner, including review questions, review chapters, or related explanation videos; the review content recommendation algorithm is optimized by the following formula:

[0128] Among them, R i,j is the recommended value of review content for the i-th learner on the j-th knowledge point, P i,j is the learner’s progress in mastering the jth knowledge point, E i,j,k is the proportion of the k1-th type of error made by the learner on the j-th knowledge point; if R i,j The higher the value, the system will provide more review content, especially for the parts with more errors. Through the above steps, personalized learning suggestions can be generated according to the learners' mastery progress and error types on each knowledge point. Learning progress adjustment, wrong question review and review content recommendation can accurately help learners strengthen weak links, improve learning efficiency, and effectively improve learning outcomes.

[0129] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0130] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. Intelligent learning behavior analysis and knowledge mastery evaluation method, characterized by: The following steps are involved: S1: Collect learners’ behavioral data during the learning process, including learning time, learning frequency, learning progress, answering speed, error rate, and click behavior; S2: Perform data cleaning and data normalization on the behavioral data collected in S1 to form a behavioral data set; S3: Extract the behavioral characteristics of learners at each learning stage from the behavioral dataset, including learners’ activity, learning time distribution, and answer accuracy; S4: Based on the behavioral characteristics of the learners, a cluster analysis algorithm is used to construct a behavior pattern model of the learners, so as to identify the behavior patterns of different types of learners; S5: Based on the identified behavior patterns and the learners’ learning performance on each knowledge point, the learners’ knowledge mastery on each knowledge point is evaluated. The specific evaluation includes the progress of knowledge point mastery, error type analysis and review frequency. S6: Generate personalized learning suggestions based on the evaluation results of knowledge mastery in S5, including learning progress adjustment, wrong question review and review content recommendation.

2. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 1 is characterized in that: The S1 specifically includes: S11: collecting the learner's learning behavior data in real time through a sensor module arranged on a learning device or a learning platform, wherein the sensor module includes a user input device, a display device and an interaction submodule, and is used to record the learner's interaction behavior during the learning process; S12: Based on the learners' interactive behavior data on the learning platform, record the learners' learning time and learning frequency. The learning time is calculated by recording the time difference between each learner's login and exit from the learning platform, and the learning frequency is calculated by recording the time interval between each learner's learning. S13: by analyzing the learning tasks selected by the learners in the learning platform, recording the learning progress, wherein the learning progress is obtained by calculating the time taken by the learners to complete the tasks at each stage and the task completion status; S14: by monitoring the learner's input response during the process of answering questions, the learner's answering speed is recorded, and the answering speed is counted by calculating the time required for the learner to answer each question; S15: recording the learner's error rate according to the errors made during the learner's answering process, wherein the error rate is calculated by calculating the ratio of the number of wrong answers to the total number of answers; S16: By analyzing the learner's click behavior, the learner's click behavior data on the platform is recorded, and the click behavior data includes the page clicked by the learner, the selected learning content and the browsing time.

3. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 1 is characterized in that: The S2 specifically includes: S21: Data cleaning is performed on the behavior data collected in S1, wherein the data cleaning includes removing missing values, removing outliers, and removing duplicate data; wherein, the processing of missing values ​​adopts the mean filling method; outliers are identified by Z-Score detection based on statistical methods, and if the Z-Score value of a data point exceeds a preset threshold, it is regarded as an outlier and is removed; duplicate data is screened by comparing timestamps and duplicates are removed; S22: Perform data normalization on the cleaned behavioral data to ensure that each data has the same dimension; S23: Classify the cleaned and normalized behavior data according to the learning stage, learning task and task completion status, and store them in a data set library to form a behavior data set.

4. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 1 is characterized in that: The S3 specifically includes: S31: According to the behavioral data set processed in S2, the learner's learning activities are processed in stages, and the learning process is divided into several learning stages, where the learning stages are defined according to different progress of the learning task or learning time interval; S32: Based on the behavior data of each divided learning stage, the activity of the learner in the stage is calculated, and the activity is obtained by counting the ratio of the learner's effective learning time to the total learning time in the stage; S33: Calculate the distribution of learners’ learning time in each learning stage. The distribution of learning time is measured by calculating the proportion of learning time in each learning stage to the total learning time. S34: extracting the correctness of the learner's answers in each learning stage, wherein the correctness of the answers is obtained by calculating the ratio of the number of answers completed by the learner in that stage to the number of correct answers.

5. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 4 is characterized in that: The S31 specifically includes: S311: According to the progress of the learning task, the learning process is divided into several stages, each stage corresponds to a different progress point of the learning task, and the progress point is defined according to the proportion of the learner completing the task. The specific calculation formula is: Among them, P task is the completion progress of the learning task, T completed is the amount of tasks completed by the learner, T total is the total workload of the task; if P task If it is greater than the preset threshold, it is divided into a new learning stage; S312: Divide the learning process into time periods according to the learning time intervals of the learners in the learning process, and each time period is represented as a learning stage; the learning time intervals are divided by setting a fixed time window, and the time window size is the average learning time of the learners. The specific calculation formula is: Among them, T window is the time window size for each learning stage, T total is the total learning time of the learner, N windows The number of learning stages to be set; S313: According to the progress of the above learning task and the result of the learning time interval division, determine the start time and end time of each learning stage. The start time and end time are determined by the intersection of the corresponding task progress and the learning time window. The specific calculation formula is: start =T previous +T window and T end =T start +T window , where T start and T end are the start time and end time of the current learning phase, T previous is the end time of the previous learning phase, T window is the time window for each learning phase.

6. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 1 is characterized in that: The S4 specifically includes: S41: Select the K-means clustering algorithm to perform cluster analysis on the learner's behavior characteristics. By setting the number of categories K and initializing K center points, the distance between each learner and the center point is calculated, and the learner is assigned to the cluster with the closest distance. This process is repeated until the clustering results converge. S42: Standardize the learners’ behavioral characteristics by using the mean-variance standardization method, adjusting the mean of each characteristic to 0 and the standard deviation to 1, to ensure that each characteristic has equal weight in the cluster analysis; S43: Through the iterative process of the K-means clustering algorithm, the position of the cluster center point is gradually optimized until the clustering results converge, and each cluster represents the behavior pattern of a learner; S44: Build a learner behavior pattern model based on the clustering results. The center point of each cluster represents a behavior pattern, which is used to describe the typical behavior characteristics of learners in the cluster, including learning activity, learning time distribution or answer accuracy rate; S45: Based on the learner behavior pattern model, different types of learners are identified by analyzing the characteristic values ​​of the clusters.

7. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 6 is characterized in that: The S41 specifically includes: S411: first set the number of categories K, and randomly select K cluster centers, wherein the number of categories K is preset according to the characteristics of the learning task and the distribution of the data; S412: Calculate the distance between each learner and the K cluster centers; S413: assigning each learner to the cluster closest to it according to the calculated distance, assigning the learner to the cluster with the smallest distance to it, and updating the cluster affiliation; S414: Update the center point position of the cluster by calculating the mean of the behavioral characteristics of all learners in each cluster to update the center point of the cluster; S415: Repeat S412 and S413 until the clustering result converges, that is, the cluster center point no longer changes significantly, or the preset maximum number of iterations is reached; S416: Through the above clustering process, learners are divided into several clusters according to their behavioral characteristics, and each cluster represents a behavioral pattern of a learner.

8. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 7 is characterized in that: The S45 specifically includes: S451: According to the learner behavior pattern model constructed in S44, the characteristic value of each cluster is obtained, including learning activity, learning time distribution and answer correctness; wherein the learning activity is calculated by the following formula: Among them, A k is the learning activity of the kth cluster, T k is the total learning time of learners in the cluster, T total is the total learning time of all learners; S452: Classify learners into different types according to the learning activity values ​​of the clusters; if the activity of learners in a cluster is higher than a predetermined threshold, the cluster represents high-activity learners; if it is lower than the predetermined threshold, the cluster represents low-activity learners; S453: Analyze the learning progress of the learners according to the learning time distribution of the cluster, and calculate the learning progress of the learners in the cluster by the following formula: Among them, P k is the learning progress of the kth cluster, T current is the learning time of the current stage, T total is the total learning time of the learner from the beginning to the present; if P k If it is greater than a predetermined threshold, the corresponding cluster is identified as a fast progress learner; S454: Based on the learner's performance in answering questions correctly, the learner type is refined; the answering correctness feature is calculated by the following formula: Among them, P correct is the correctness of the answer of the kth cluster, C k is the number of correct answers given by learners in the cluster, Q k is the total number of answers of learners in the cluster; if P correct If it is higher than the set threshold, the corresponding family represents high-accuracy learners, and if it is lower than the set threshold, it represents low-accuracy learners; S455: Through steps S451-S454, the activity, learning progress and answer accuracy of the learners in the cluster are comprehensively analyzed, and the learner types are divided into high-activity learners, low-activity learners, fast-progress learners, slow-progress learners, high-accuracy learners or low-accuracy learners.

9. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 1 is characterized in that: The S5 specifically includes: S51: According to the learner behavior pattern identified in S45, the learning performance data of each learner on each knowledge point is obtained, and the learning performance is defined as the correct answer rate of each knowledge point; S52: Evaluate the learner's progress in mastering each knowledge point based on the correct rate of the learner's answers to each knowledge point; calculate using the following formula: Among them, P i,j is the progress of the i-th learner in mastering the j-th knowledge point, C i,j is the correct answer rate of the i-th learner on the j-th knowledge point, T j is the maximum score of this knowledge point; if P i,j If the value is greater than the predetermined threshold, it means that the learner has mastered the knowledge point. If it is lower than the predetermined threshold, it means that the learner has not yet mastered the knowledge point. S53: According to the number of wrong answers of each learner on each knowledge point, error type analysis is conducted. Error types include selection errors, calculation errors or concept comprehension errors. Specifically, error type analysis is determined by calculating the proportion of each type of error. If the proportion of selection errors is greater than 30%, it is considered that the learner of this knowledge point has difficulty with this type of error and needs to focus on reviewing. S54: Calculate the learner's review frequency based on the learner's mastery progress on each knowledge point and the error type analysis results; the review frequency is determined by the learner's mastery of a certain knowledge point and the error type analysis results; specifically, if the learner's mastery progress on a certain knowledge point is lower than a preset value, and the error type analysis results show that the error ratio is greater than 30%, then increase the review frequency of the corresponding knowledge point.

10. The intelligent learning behavior analysis and knowledge mastery evaluation method according to claim 1, characterized in that: The S6 specifically includes: S61: Adjust the learner's learning progress according to the analysis results of each learner's mastery progress and error types on each knowledge point; specifically, if the learner's mastery progress on a certain knowledge point is lower than a predetermined threshold, it will be recommended to increase the learning time of the corresponding knowledge point or arrange additional learning tasks; the learning progress adjustment formula is: S62: Generate wrong question review suggestions based on the error type analysis results in S53; the wrong question review suggestions are generated by the following steps: Step 1: Sort by error type, reviewing high-frequency error types first; Step 2: Recommend corresponding exercises or analysis videos for the most common error types; S63: Based on the learner's progress in mastering each knowledge point and the analysis of the error types, review content recommendations are generated, including review questions, review chapters or related explanation videos.

Images